Innosilicon Grin Miner G32 Preliminary Specification

[Taek]

@papacabeza

But Obelisk stands out by not having delivered any single delivery unit to specification or on time. The company also stands out for making promises for refunds and then changing it’s mind later. Finally, the company stands out by telling a court that it’s DCR sales were a “Kickstarter Campaign.” (recording here

Obelisk has only ever shipped one line of products - the SC1 and DCR1 chip, both of which were on the same reticle. Foundry based delays hit both the same way because… they are on the same wafer. We made up for it by putting forward a compensation policy which was completely fair - we gave our users credits equivalent to the revenue that they would have received from mining had their machines arrived on time. If any other user besides papacabeza feels that this compensation policy is not completely fair, I would love to hear about it.

@papacabeza

but by contract you’ve required your customers to give up all rights to any late delivery or tech specifications and Obelisk has absolved itself of all liability.

Certain things are outside of our control. For example, TSMC has occasionally experienced delays due to things like tsunamis. If that were to happen, Obelisk would be unable to avoid missing a shipping date, because as a startup we don’t have the capital to source our chips from two different manufacturers, and we also don’t have the establishment to buy insurance.

Innosilicon also doesn’t have this type of protection, and in those events they would not be able to follow through on their guaranteed refund. For example, Innosilicon offered their customers ROI protection for a while on orders of T2 units. To the best of my knowledge, these units never did ROI, and also Innosilicon never paid out on the insurance they gave their customers.

In short: Innosilicon is making guarantees that they wouldn’t be able to follow through on, and they have done this in the past, and Obelisk is not willing to do that.

If we were a larger company that could afford to write off $20 million because TSMC got hit by a natural disaster and delivered chips 3 months late, we would give our users guarantees or insurance. We can’t do that though, therefore we can’t make promises that way.

[gary]

I didn’t find the yield data. How many G32 do you plan to offer? or where to track the pre-order numbers?

Again! Please Innosillicon provide these key numbers.

A miner sales with unknown total number doesn’t make sense, we can’t estimate the ROI data if we don’t know the total number.

[papacabeza]

@Taek

Foundry based delays hit both the same way because… they are on the same wafer. We made up for it by putting forward a compensation policy which was completely fair - we gave our users credits equivalent to the revenue that they would have received from mining had their machines arrived on time.

The foundry selection, contract, timing, management and contingency plans (if any) are entirely under Obelisk’s control. Your contract shifted the entire liability for that risk on to customers. As a result. I and the other customers paid in full for our miners — mine arrived a year later as bricks. You shifted all risk in the whole production chain on to the customer. That’s your business model.

You keep talking about this compensation plan. Can I still participate? I would sure like something out of my $20k. How can I cash in on that now, please?

[papacabeza]

@Taek

One thing that would be pretty amazing is if Obelisk, Innosilicon, and other manufacturers could come to some sort of mutual agreement for moving forward in a way that ensures everyone comes out ahead.

I don’t see that happening, but we’re more than willing to discuss.

I believe, though I can’t be certain, that Obelisk has the lowest cost basis for producing hashrate. As long as NRE can be covered, Obelisk is willing to supply chips to everyone, including Innosilicon, under some open agreement. We’re also willing to consider alternative routes forward.

You might find this interesting reading, David: the Clayton Act of 1914.

[pinoli]

Has anyone noticed this little clause in the Innosilicon G32 miner sale page?

You are eligible for full refund if our final G32 spec after actual product testing fails to meet the above Performance Target by Sept 30th this year

Not only, this applies exclusively to April buyers:
Guarantee for April pre-order customers only:

This smells fishy, they claim they will ship in August without specifying a single date in the month: I am to assume they will ship the last useful day, so basically September. The clause strongly indicates they know production of the initial units will be finished around mid September at most.

Now Obelisk claims they will ship GRNx miners in October, also without specifying a date. But, given past performance, I am quite sure they are being extremely conservatives with their estimates. I think @Taek (probably in this thread) basically confirmed this assumption.

I truly believe it will be a race, they are both aware they are really close and I am sure shipping of both miners will overlap for some time.

I also believe that once all miners are online, the G32-Mini will be kind of useless (and I guess this has been the most sold model, possibily responsible of the biggest portion of Inno hashrate on Grin chain at launch). They are also 25% less efficient than GRN1 on a Watt per GPS basis, so I don’t know how long they will stay relevant (considering Inno and Obelisk would start selling new batches in the meantime).

I also fail to see how Inno price structure makes sense: they sell their best miner (G32-1800), able to compute 328 GPS @ 1800 W, for 10.000$ (tomorrow it will go for 15.000$) while Obelisk sells a more powerful miner (GRN1, 420 GPS @ 2200 W, slightly more efficient +5%) for 3.000$/10.000$ (the latter from May 3rd).

Also, I fail to see how Inno could flood the market with these prices: as of now Obelisk sold 137 GRNx units total. I can believe Inno could have higher numbers since its more established and well known, but I don’t think these (both Inno and Obelisk) will be very popular miners (given also the fact they will be useless by August 2020).

Bottom line, after all this drama fades out I am sure we will be back to the boring stuff: efficiency, reliability, maximum performance. If this holds, then I can see Obelisk clearly on top, they have the best offering hands down.

Just my two cents, what do you think? If you spot any inaccuracies let me know.

[randomdude]

You assume the worst from Inno, an established player, and best from Obelisk, a player whose past record speaks for itself. Very unlikely. I would guess Inno ships early August and Obelisk mid-September.

Note that Obelisk has committed to 2500 units (equivalent), so they are not going to produce another batch for C31. Inno is a different story.

Almost everything depends upon the future price of Grin:

• If Grin does not move much, then it all comes down to how many units Inno will produce. If they produce as much hash as Obelisk units do in total, then both side buyers will lose.

• If the price does increase, initial buyers will make money. But then Inno will pump out more units, at lower prices, reducing everyone’s profitability.

In either case Inno will end up capping anyone’s gains. So Inno holds all the cards.

Don’t go by Obelisk pre-order numbers, almost everyone is smarting from previous Obelisk experience.

And I don’t think power consumption would be material given the short life cycle C31.

[pinoli]

@randomdude

You assume the worst from Inno, an established player, and best from Obelisk, a player whose past record speaks for itself

basically yes for two reason:

1. given past Obelisk screwup I highly doubt they can afford to go through with another debacle. That is why I am sure they will surprise us, because if they fail I think it’d get really ugly for Obelisk.
2. Inno move, to my eyes, was just a way to counter Obelisk. Their initial launch on Twitter seemed sketchy and rushed at best, available miners are subpar, and now weird clauses pop up: from all of this, I can safely assume Inno improvised and it’s making it along the way.

@randomdude

Note that Obelisk has committed to 2500 units (equivalent), so they are not going to produce another batch for C31. Inno is a different story.

I agree Inno could flood the market, but I don’t think it’s in their interest. If Grin works, Inno will be producing Grin miners for a long time, while if the project is killed by a sudden flood of miners Inno will lose long term revenues on the project. Even if Grin survives, after C31 fades out people are going to remeber who scammed them up with unprofitable miners and will think twice when purchasing C32 miners.

Reputation is key here.

@randomdude

Don’t go by Obelisk pre-order numbers, almost everyone is smarting from previous Obelisk experience.

yeah I am sure numbers will shoot up after the sale ends. But I am still convinced we are not going to see 10.000 miners on the Grin network in October 2019. Grin, as much as I love it, it’s still very small and experimental and not very well known outside these niche communities. What fraction of this already small userbase is willing (and able) to spend tens or hundreds thousand dollars? Also, if sales reach the 5 digit range this means almost 100 M$ has been spent on Grin hardware.
It would awesome and flattering, but i highly doubt it at least not for C31, I don’t see it.

@randomdude

And I don’t think power consumption would be material given the short life cycle C31.

this sounds right: given the short life span of the miners maximum performance will be much more important than overall efficiency, but still, Obelisk has higher maximum speed. So if I had to buy a miner, I would go for a GRN1 rather than a G32-1800, since If they ship around the same time I would make more Grin with the Obelisk unit. Plus it would cost less.

What I mean is that Inno can do whatever they want, but they still need to find buyers as any other company. And their offering is poor, while Obelisk proposal is more clear cut and with higher performance and cheaper price. So buyers will definitely do their math and see if Inno miners are worth it, given their insanely high prices and subpar performances.

Regarding the price, it really depends: if you sell as soon as you mine it, price can be relevant, but if one’s goal is to hoard Grin and hodl it doesn’t make any difference.

Let’s see how this pan out.

[papacabeza]

@pinoli

Now Obelisk claims they will ship GRNx miners in October, also without specifying a date. But, given past performance, I am quite sure they are being extremely conservatives with their estimates. I think @Taek (probably in this thread) basically confirmed this assumption.

To the contrary, Obelisk has confirmed that they are not changing any of their contract terms, and that they are shifting 100% of all delivery risk and technical specifications on to the consumer. It’s true that they are saying that the estimates are conservative, but the contract says just the opposite.

[papacabeza]

@pinoli

given past Obelisk screwup I highly doubt they can afford to go through with another debacle. That is why I am sure they will surprise us, because if they fail I think it’d get really ugly for Obelisk.

Respectfully, your assumption is wrong. Obelisk has actually done very well for itself in the first two products, SC1 and DCR1, making more than 30m in revenue for the company. They have nothing to lose by trying to raise more money by these presales — the company takes on zero risk and shifts it all to consumers.

[pinoli]

@papacabeza

Respectfully, your assumption is wrong. Obelisk has actually done very well for itself in the first two products, SC1 and DCR1, making more than 30m in revenue for the company

sorry I was referring to their poor performance with customers, not with revenues: I still believe they cannot really gamble on the next delivery for reputational reasons.

@papacabeza

the company takes on zero risk and shifts it all to consumers.

I disagree when you say “zero risk”: there is no such thing as zero risk in business and I guess at a point they had some invoice to settle, like several millions toward the manufacturer.
plus screwing up hardware designs is easier than you think, the gamble they took (as Inno btw) is enormous.

on a lighter note, I really think ANYONE involved with experimental mining machines must be a bit crazy. Manufacturers, customers, pools and so on, they all show courage by wasting insane amount of time and money toward securing a specific network. if you see it from a broader perspective, this process is novel, unstoppable and ultimately amazing. This is to say that I don’t see customers (I am one) as “poor innocent lambs” who get exploited while the companies reap profits taking “zero risks”.
This is all a gamble, for many the gambe of their lifetime, and one should be ready to face the consequences and, possibly, ultimate failure.

[papacabeza]

@pinoli

I disagree when you say “zero risk”: there is no such thing as zero risk in business and I guess at a point they had some invoice to settle, like several millions toward the manufacturer.

This isn’t opinion, it’s fact — the contract shifts all risk of delivery on to customers. Of course the company still has to pay the manufacturer — that’s not risk, that’s cost of goods.

[pinoli]

@papacabeza

This isn’t opinion, it’s fact

dont wanna argue, but if it’s a fact than we all should start mining hardware companies and make billions “risk free”
I still disagree, as I stated above they still have to sell: if they don’t sell enough they go under.
Yes, they could have leftover revenues from SC1 and DCR1 to cover for poor sales in this phase, but this is another gamble they take. It’s how business works, it’s a continuous enterprise with continuously changing risks/rewards (what if there is a tsunami at the fab, what if we screwed up the mask and wasted 5 M$ , what if our warehouse went on fire the day before delivery, what if Grin shooted up to 100$ and so on…).

anyway, I really hope they ship at the same time lol
i would like to see my theories tested in the field

[papacabeza]

@papacabeza

Foundry based delays hit both the same way because… they are on the same wafer. We made up for it by putting forward a compensation policy which was completely fair - we gave our users credits equivalent to the revenue that they would have received from mining had their machines arrived on time.

[ . . . ]

You keep talking about this compensation plan. Can I still participate? I would sure like something out of my $20k. How can I cash in on that now, please?

@Taek coming back to this. The compensation plan that you keep mentioning. Please let me know what I can do to participate so that my $20k loss isn’t a total loss. How can I get the compensation please?

[Taek]

@papacabeza

Respectfully, your assumption is wrong. Obelisk has actually done very well for itself in the first two products, SC1 and DCR1, making more than 30m in revenue for the company. They have nothing to lose by trying to raise more money by these presales — the company takes on zero risk and shifts it all to consumers.

To the contrary, Obelisk ate a loss on the DCR1, and your $30mm revenue estimate is much higher than our actual revenue. Did you account for coupons? Did you account for the lower prices that we sold the units at throughout the sale? Your estimates around our expenses are also likely far too low. Obelisk takes risks alongside its customers.

To assert that Obelisk has zero risk from being in business is completely absurd.

@papacabeza

@Taek coming back to this. The compensation plan that you keep mentioning. Please let me know what I can do to participate so that my $20k loss isn’t a total loss. How can I get the compensation please?

If you bought your units from Obelisk directly, you will see that your account has been credited with some Obelisk credits. Unlike coupons, these credits can be applied to orders with no limits and stacked with coupons or other credits. Obelisk currently does not have enough money to pay out the credits in cash or btc, however as soon as Obelisk has enough money to be able to pay out the credits without jepordizing our business, we will allow the credits to be withdrawn as BTC as well.

We would like to be able to pay out the credits now, but that would immediately bankrupt us, and everyone would only receive a fraction of what we would like to give them. So we will have to wait until we are sufficiently profitable as an entity to make the payouts.

[jbit2020]

@Taek Hi Are you sure about that? ASIC miners are not profitable? please explain me to stop ordering G32 miner
thanks

[imegi]

it’s fine, they were just contemplating about ROI and the trustworthy-ness of oberlisk

[Mr36]

Do you mind explaining how you calculate to get to 5000 units x 328 GPS= 1,640,000. That much I understand. Then at $3 per GRIN you say one machine makes $6,624 over one year. Is that for the whole year, or per month or what? How do you get to that figure?

[randomdude]

That came from the total amount of GRIN to be emitted until phase-out in Aug-20 for C31. There was a chart somewhere…

[UncleWiggly]

Have people given up on this discussion? Has anyone who bought from any of the so-called “Trusted Vendors” listed on the “www.asicminervalue” website gotten what they ordered, or gotten a refund for any pre-payment? Isn’t there an international fraud prevention agency that could launch an action against Innocilicon or Obelisk? What about a class action lawsuit? Has global finance been paying these companies to slow-walk production? Not hard to speculate that any government/bankster agency who saw the profitability graphs of these machines could have said, “This has got to be stopped!” We all know now that regulators and Wall Street have been manipulating BTC at least since December 2017. --Any thoughts? Positive or negative…

[LokarBoA]

I ordered direct from Innosilicon on Feb 27th and the most recent shipment date I’ve been given is January 2020. I’ve seen Inno state that they will offer refunds but I haven;t seen any follow-up to that where you could actually get one. If you bought from a 3rd party I think I’d be trying to get them to refund you but I don’t know of any groups or anything that are pursuing that.